Q_('ctxt|msg') expands to g_dpgettext(NULL, 'ctxt|msg',0) and should be
interpreted as a message with message context. According to
https://www.gnu.org/software/gettext/manual/gettext.html#Language-specific-options
this glib extension is supported using the 'g' suffix to the keyword spec.
This was missing in our xgettext invocation.
tbl-width is not necessarily an even number; tbl-width being odd would
result in a half-fraction when calculating number of
empty-cells. convert to the appropriate integer.
this is the proper fix for the bug fixed by d865b149.
assoc-ref (or better, assq-ref because we're comparing symbols) is
used to lookup from the 'car's of a list of pairs, and return the
found pair's 'cdr'. the previous use of 2-element lists demonstrate a
lack of understanding how to encode the list of pairs. rewrite using
proper pairs rather than 2-element lists, which means the assq-ref
does not need a subsequent 'car' to retrieve the desired symbol.
* this function has a typo in name
* all uses of accounts-get-all-subaccounts were followed by appending
the result to the original accounts list. we have already rewritten to
use the better function in previous commit. this is now obsolete.
* inline its last use, omit sorting. list is sorted anyway afterwards.
All uses of gnc:acccounts-get-all-subaccounts were immediately
followed by appending the result to the original accounts list. Use
gnc:accounts-and-all-descendants instead which is more efficient.
When the ledger is loaded, the list for the taxtable combo is created
but as the item_list gui is not initialised yet a temporary list store
is used. This is done twice and hence end up with duplicate entries. To
fix this add the missing gtk_list_store_clear to
gnc_combo_cell_clear_menu to clear the temporary list store when the
item_list is not initialised.
When you start the import process, possible imported transactions are
displayed in the source open register and if you deselect any and press
the OK button you are left with blank lines in the register for those
deselected so move the gnc_gen_trans_list_delete to before the gui
refresh call.
this leads to better error messages during test development and other
work. e.g. a report-module isn't loaded by mistake during a test,
would formerly lead to a cryptic #f error; this commit ensures the
uuid cannot be found error is reported.
* the calc_clicked_cb had mechanism to abort if no fields were
empty. disable it.
* to convert gnc_numeric to string, the only suitable print_info is
gnc_share_print_info_places which takes decimal and outputs suitable
print_info. Not very appropriate, but works.
Previously, for general-journal, each split was being analyzed, and
only the source split was counted for totals. This change will means
the source split *and* its peers were accounted for.
We don't recommend this as it risks tainted build environments. However if
you know what you're doing it should still work. In particular sandboxed systems
such as flatpak don't have this problem and hence they do build in-tree.
* Use cmake's built-in copy command rather than an external one
* depend directly on gnc-vcs-info.h rather than the global gnc-vcs-info-target
The latter won't trigger a rerun of the command if gnc-vcs-info.h already exists
regardless of whether it had changed.
